search

jbloomAus / SAEDashboard

MIT License
18 stars 3 forks source link

fix: fixing matmul error when using non float32 dtype #1

Closed chanind closed 4 months ago

chanind commented 5 months ago

This PR fixes a bug where tensor dtypes don't match when settings a dtype to anything other than float32.

jbloomAus commented 4 months ago

@chanind sorry I missed this. Should work now as I likely saw the same issue. Can you check and let me know?

jbloomAus commented 4 months ago

@chanind pinging you about this.

chanind commented 4 months ago

Ack sorry, closing!